Changes

[lldb] Rename NativeRegisterContext{Watchpoint => DBReg}_x86

Differential Revision: https://reviews.llvm.org/D97210

[lldb] [test] Workaround symlink-related test failures

Use realpath() when spawning the executable create_after_attach
to workaround a FreeBSD plugin (and possibly others) problem.
If the executable is started via a path containing a symlink, it is
added to the module list twice -- via the real and apparent path.
This in turn cases the requested breakpoint to resolve twice.

Use realpath() for main program path in lldb-vscode breakpoint tests
to workaround a similar problem.  If the passed path does not match
the realpath, lldb-vscode does not report the breakpoints as verified
and causes tests to fail.

Since the underlying problems are non-trivial to fix and the purpose
of these tests is not to reproduce symlink problems, let's apply
trivial workarounds to make them pass.

Differential Revision: https://reviews.llvm.org/D97230

[lldb] [test] Skip AVX lldb-server test on non-x86 architectures

Skip the AVX-related lldb-server test on non-x86 architectures, as they
do not support AVX.  While technically the test worked on Linux because
the AVX check would simply return false, other platforms do not provide
such a straightforward way of checking for AVX (especially remotely),
and the results of such check may need to be interpreted specially
for the platform in question.

Differential Revision: https://reviews.llvm.org/D97450

[PowerPC] Removing sqrtd2 and sqrtf4 from list of vectorizable function with MASSV

Under -O3 and -Ofast, the MASSV conversion prevents the sqrt call to be inlined.
Inline sqrt is faster than MASSV call on leppc.

Differential Revision: https://reviews.llvm.org/D97487

[TableGen] Avoid repeated TreePredicateFn::getCodeToRunOnSDNode() calls in MatcherTableEmitter::EmitNodePredicatesFunction loop. NFCI.

[WebAssembly] call_indirect issues table number relocs

If the reference-types feature is enabled, call_indirect will explicitly
reference its corresponding function table via TABLE_NUMBER
relocations against a table symbol.

Also, as before, address-taken functions can also cause the function
table to be created, only with reference-types they additionally cause a
symbol table entry to be emitted.

Differential Revision: https://reviews.llvm.org/D90948

[VPlan] Remove recipes from back to front.

Update the deletion order when destroying VPBasicBlocks. This ensures
recipes that depend on earlier ones in the block are removed first.
Otherwise this may cause issues when recipes have remaining users later
in the block.

[clang-cl] make -f(no-)ident a CoreOption

On clang emits the compiler version string into debug information
by default for both dwarf and codeview. That makes compiler output
needlessly compiler-version-dependent which makes e.g. comparing
object file outputs during a bisect hard. So it's nice if there's
an easy way to turn this off.

(On ELF, this flag also controls the .comment section, but that
part is ELF-only. The debug-info bit isn't.)

Differential Revision: https://reviews.llvm.org/D97695

[clang] SimpleMFlag helper in Options.td

This is the new helper to create a boolean -m and -mno-
options.

Differential Revision: https://reviews.llvm.org/D97069

[libc++/abi] Allow running back-deployment testing against libc++abi

Before this patch, we could only link against the back-deployment libc++abi
dylib. This patch allows linking against the just-built libc++abi, but
running against the back-deployment one -- just like we do for libc++.

Also, add XFAIL markup to flag expected errors.

Differential Revision: https://reviews.llvm.org/D91069

[TTI] Consider select form of and/or i1 as having arithmetic cost

This is a patch that updates the cost of `select i1 a, b, false` to be equivalent to that of `and i1 a, b`
as well as the cost of `select i1 a, true, b` equivalent to `or i1 a, b`.

Until now, these selects were folded into and/or i1 by InstCombine, but the transformation is poison-unsafe.
This is a step towards removing the unsafe transformation. D93065 has relevant transformations linked.
These selects should be translated into the assemblies as and/or i1 do in the same manner. The cost should be equivalent.

Reviewed By: spatel

Differential Revision: https://reviews.llvm.org/D97360

Fix a warning about named return value not being moved-from.

The use of an rvalue reference here was using reference lifetime
extension needlessly - the code is simpler and more efficient without
it.

Fix virtual-dtor warning a different way, since a virtual dtor is not required for this use case.

[lld-macho][nfc] Remove TODO regarding addends

There was initially some concern around the correct handling of pcrel
section relocations with r_length != 2. But it looks like there are no such
relocations in practice -- x86_64's pcrel section relocs all have r_length == 2,
and ARM64 doesn't even have pcrel section relocs. So we can replace the TODO
with an assert.

Reviewed By: #lld-macho, thakis

Differential Revision: https://reviews.llvm.org/D97576

[lld-macho] Switch default to new Darwin backend

The new Darwin backend for LLD is now able to link reasonably large
real-world programs on x86_64. For instance, we have achieved
self-hosting for the X86_64 target, where all LLD tests pass when
building lld with itself on macOS. As such, we would like to make it the
default back-end.

The new port is now named `ld64.lld`, and the old port remains
accessible as `ld64.lld.darwinold`

This [annoucement email][1] has some context. (But note that, unlike
what the email says, we are no longer doing this as part of the LLVM 12
branch cut -- instead we will go into LLVM 13.)

Numerous mechanical test changes were required to make this change; in
the interest of creating something that's reviewable on Phabricator,
I've split out the boring changes into a separate diff (D95905). I plan to
merge its contents with those in this diff before landing.

(@gkm made the original draft of this diff, and he has agreed to let me
take over.)

[1]: https://lists.llvm.org/pipermail/llvm-dev/2021-January/147665.html

Reviewed By: #lld-macho, thakis

Differential Revision: https://reviews.llvm.org/D95204

Remove use of tuple for multiresult type storage

Move the results in line with the op instead. This results in each
operation having its own types recorded vs single tuple type, but comes
at benefit that every mutation doesn't incurs uniquing. Ran into cases
where updating result type of operation led to very large memory usage.

Differential Revision: https://reviews.llvm.org/D97652

[AMDGPU] Add IntrWillReturn to recently added intrinsics

This adds IntrWillReturn to the gfx90a mfma intrinsics, to match all the
other mfma intrinsics, and llvm.amdgcn.live.mask, to match
llvm.amdgcn.ps.live.

Differential Revision: https://reviews.llvm.org/D97675

[clangd] Improve document symbols support for Objective-C categories and methods

- Categories will now show up as `MyClass(Category)` instead of
  `Category` and `MyCategory()` instead of `(anonymous)` in document
  symbols

- Methods will now be shown as `-selector:` or `+selector:`
  instead of `selector:` to differentiate between instance and class
  methods in document symbols

Differential Revision: https://reviews.llvm.org/D96612